Stellalpha

Autonomous, non-custodial capital allocation infrastructure for Solana.

App · Whitepaper · DoraHacks · Vault Repo · X

Stellalpha is a non-custodial execution layer designed to replicate trade intent instead of raw transaction data. By moving beyond naive wallet mirroring, Stellalpha enables stablecoin capital to follow high-signal on-chain traders with institutional-grade risk alignment and sub-second execution.

The project is the result of 6 months of heavy engineering focused on solving the "balance desynchronization" problem—where traditional copy-bots cause followers to take disproportionate risk compared to the leader. Stellalpha's engine interprets a trader's allocation weight and executes a proportional trade within the follower's secure vault environment.

---

🛠 Technical Moat: The Execution Engine

Stellalpha is engineered for professional reliability, solving the latency and parsing challenges that kill traditional copy-trading platforms.

• High-Performance Rust Parser: Our custom carbon-parser (built on the Carbon framework) decodes complex SVM instructions at the source, normalizing trades from Jupiter, Raydium, and Meteora into a unified "Intent Model."
• Dual-Ingestion Pipeline: Powered by Yellowstone gRPC for ~200ms signature detection and Helius Enhanced Webhooks for high-fidelity data reconciliation.
• Atomic Deduplication: A Redis-backed control plane ensures that only the fastest ingestion source "claims" a trade for execution, preventing costly double-trades.
• Adaptive Allocation Models: Instead of replaying fixed dollar amounts, we replicate Allocation Weight. If a leader sells 10% of their SOL, Stellalpha sells 10% of the follower's SOL, preserving strategy fidelity across any wallet size.

---

🧱 The Infrastructure Stack

• Core Logic: Rust (Carbon-core), TypeScript (Next.js 15, React 19)
• Data Ingestion: Yellowstone gRPC (Chainstack/PublicNode), Helius Webhooks
• DeFi Execution: Jupiter SDK & API (Routing, Quotes, and Simulation)
• Control Plane: Redis (Upstash/Railway), Supabase (PostgreSQL & Auth)
• Deployment: Railway (Distributed Ingestion Workers), Docker

---

The Thesis

Most copy trading systems are built around wallet mirroring. That works in a lab, but it breaks down in production:

• Portfolio Drift: Follower balances never perfectly match leaders.
• Execution Latency: 2-second delays cause massive slippage on high-velocity tokens.
• Accounting Mess: Replaying 100 small trades creates complex tax and tracking nightmares.

Stellalpha's model is simple: Replicate the trade decision, not the raw transaction.

By using stablecoin capital as the strategy base and replicating intent-weights, we ensure clearer follower accounting and a robust path to institution-ready strategy infrastructure.

---

What Is Live Today

• Intent-Based Simulation Engine: A high-fidelity environment validated over 6 months that uses real-time market data to prove the weight-based replication thesis.
• Star Trader Discovery: Performance-focused leaderboard for browsing curated Solana wallets and their historical "Intent Patterns."
• Real-time Ingestion Stack: Yellowstone gRPC backend provides sub-second trade detection.
• Trader Detail Surfaces: Deep-dive analytics into wallet-level holdings and follower-facing PnL context.
• Stale-Signal Protection: Asymmetric execution policy that skips BUYs older than 10s but ensures SELLs are always fulfilled to protect exits.

---

Real-Time Ingestion Path

flowchart LR
    A["Tracked wallets"]

    subgraph S1["Detection"]
        B["Yellowstone gRPC"]
        C["Webhooks"]
        D["Reconciliation"]
    end

    subgraph S2["Transaction Processing"]
        E["Durable queue"]
        F["Primary parsing"]
        G["Fallback parsing"]
        H["Trade detection"]
    end

    subgraph S3["Decision Engine"]
        I["Atomic deduplication"]
        J["Follower allocation"]
    end

    subgraph S4["Execution & State"]
        K["Demo vault simulation"]
        L["Portfolio, PnL, and trade history"]
        M["Future mainnet vault execution"]
    end

    A --> B
    A --> C
    A --> D
    B --> E
    D --> E
    E --> F
    F -->|unresolved only| G
    F --> H
    G --> H
    C --> H
    H --> I
    I --> J
    J --> K
    K --> L
    J --> M

Loading

---

Mainnet Vault Integration

Stellalpha's intended mainnet architecture is centered on stellalpha_vault, an on-chain layer built using Anchor to turn simulated trades into real, non-custodial execution.

The model uses a Vault -> Trader State architecture:

• Users own a base vault that holds their USDC capital.
• Following a trader creates a dedicated Trader State PDA.
• Stellalpha receives delegated execution authority, constrained to specific policy envelopes (e.g., "Jupiter-only," "Max 5% slippage").

---

🏁 Technical Verification Guide

To verify the Intent-Based engine in action:

1. Connect Wallet: Sign in to stellalpha.xyz via Phantom or Solflare.
2. Create Demo Vault: This initializes 1,000 virtual USDC for strategy testing.
3. Follow a Star Trader: Select a high-activity wallet from the leaderboard.
4. Monitor Execution: Watch the dashboard as the gRPC engine detects trades and calculates the Copy Ratio based on your vault's relative size.
5. Code Audit: Inspect the carbon-parser/ directory for the Rust decoding logic and lib/copy-models/ for the allocation math.

---

Built in public for the Solana ecosystem.